Description of the Method Developed for Dye Penetrant

Analysis of Cracked Solder Joints

Background
The extension of cracks in solder joints after fatigue testing is usually evaluated using cross-
sectioning of the solder joints. The printed board assembly, or a part of it, is potted in epoxy
and is then cut close to the solder joints that will be examined and the sample is ground in to
the solder joints. As shown in Figure 1, the formation of cracks in solder joints can be very
irregular and vary considerably even in joints located close to each other.

Figure 1 Photograph showing the extent an irregularity of cracking in solder joints to a

BGA after thermal cycling. The cracking has been analysed using a dye
penetrant and the areas coloured red indicate how far the cracks had
propagated during the thermal test.
Thus, to obtain a three-dimensional view of the extension of cracking in a solder joint,
polishing must be done in steps while the extension of cracking is successively documented.
Just making one cut through a solder joint may give a completing misleading result. For
example, if a cut is made from A to B in Figure 1, it would indicate a completely cracked joint
whereas a cut from C to D would indicate that almost no cracking has occurred. Furthermore,
a number of solder joints need to be examined in order to get statistically valid results. All
solder joints in a row can be examined simultaneously. If all solder joints to a BGA shall be
examined, this must be repeated for every row of joints. This is a very time consuming and
costly process. Furthermore, it is difficult to determine the exact position of a cut.
As already shown in Figure 1, dye penetrant analysis offers a three-dimensional view of
cracks in solder joints to a component. The technique was originally developed by Motorola’s
Land Mobile Products Sector’s Advanced Manufacturing Technology group [1]. The
component assembled on the printed circuit board was flooded with a dye penetrant liquid to
define the cracked area. After having applied the dye, the component was removed by
bending the board numerous times. The fractured solder joints were then visually inspected
and photographs were made to record the position that dye had penetrated into the joint. This
method works well if the cracks in the solder joints are large. If they are not large enough, the
pads on the board will rip off from the laminate.

1
In order to facilitate the removal of components with small cracks in the solder joints, IVF has
improved the technique to remove the components in a project funded by ESA [2].
Application of Dye
In order to facilitate the application of the dye, a dam is created around the component using a
modelling wax. If there are via holes beneath the component, a tape is applied to the via holes
on the opposite side of the board to prevent the dye from escaping through them. If that is not
possible, the sample is laid in a glass beaker. The dam is then filled with the dye (Steel Red
from DYKEM) or alternatively the dye is poured into the glass beaker until the sample is
completely submerged in the dye. The board or the glass beaker is placed in a vacuum
chamber and two evacuations are made down to 100 mbar in order to remove any air
entrapped in the cracks. Since solvents in the dye evaporate during this process, it is done as
fast as possible to prevent the viscosity of the dye getting too high. The surplus dye is then
poured out and the sample is dried at 100°C for 15 minutes.
Removal of Components
The upper side of the component package is roughened using a grinding paper and is then
dried with a cloth wetted with acetone. A steel cylinder with a threaded hole is glued to the
component using a two-part epoxy glue, Plastic Padding Super Steel from Loctite (Fig. 2).
The steel cylinder is shotblasted in order to improve the adhesion of the glue.
The board is then fixed by screws to an aluminium plate with a thickness of 1 cm. A hook is
screwed to the steel cylinder and the sample is arranged so that a pulling force can be applied
through the hook (Fig. 3). For BGA components having only peripheral joints it may be better
to drill a hole through the component and apply a pulling force as shown in Figure 4. The
aluminium plate is placed on a heating plate with the heat controlled by a thermocouple
attached to the upper side of the board. The temperatures of the solder joints are registered
using another thermocouple.
A pulling force of about 50 grams per joint is applied to the package. This force causes the
solder to creep and the package can be removed without bending the board. If the solder joints
are severely cracked, it may be possible to remove the package within a few hours at room
temperature. However, in most cases it is necessary to heat the samples to be able to remove
the packages within a reasonable time. The higher temperature, the faster is it possible to
remove the component. By heating the solder joints to 140°C, even packages with only small
cracks in corner joints can be removed within a few hours. However, at temperatures above
120°, the colour of the dye fades slightly. Also, a good temperature control system is required
to ensure that the temperature of the board will not rise above the melting point of the solder
when the package becomes loose. If the maximum temperature is limited to 120°C, it may
take 10-40 hours to remove a BGA component having about 600 I/Os if the cracks are very
small in the solder joints.

2
Figure 2. A component with a steel cylinder glued to it

Figure 3. Arrangement for applying a pulling force to a package

3
Figure 1 Arrangement for applying a pulling force to a BGA by drilling a hole through the
component.

Case Study of DBGA

This example taken from Reference 2 shows how dye penetrant analysis has been used to
evaluate the extent of cracking in solder joints to two thermally cycled Dimple BGA (DBGA)
packages.
Test Vehicle
The DBGA packages had a body size of 17 x 17 mm and 228 I/Os with a pitch of 1.0 mm.
Both packages had daisy chain interconnections. The packages were soldered to a multilayer
polyimide/glass board with tin-lead plated pads using vapour phase soldering. A flux was
added to promote wetting but no additional solder. One package was soldered to a footprint
with round solder pads (Component A) whereas the second package was soldered to a
footprint having pads with a teardrop form (Component CX).
The test vehicle had been thermally cycled in air per ESA-PSS-01-738 with temperature
extremes of –55 and +125°C. The cycling was stopped after 500 cycles. The package soldered
to the footprint with round pads still functioned electrically after the thermal cycling test but
one electrical defect had occurred in one row of I/Os on the other package.
Dye Penetrant Analysis
The extent of cracking in the solder joints to the two DBGA packages was analysed using the
procedure for dye penetrant analysis described previously. The solder joints were heated to
about 100°C during the removal of the components. At this temperature, the components were
removed within 30 minutes.
All solder balls remained on the printed circuit board for both packages. The fractures in the
solder joints were very close to the component pads for all solder joints (Fig. 5). The majority
of the solder joints were severely cracked for both components and to about the same extent
(Fig. 6). The fractures in some corner solder joints were coloured completely red (Fig. 7).
Thus, although failure had only been registered for Component CX, failure for Component A
must have been imminent.

4
Figure 5. View of solder balls remaining on the PCB after removal of Component CX

Figure 6. Extent of cracking in Component A (a) and to Component CX (b) analysed using
dye penetrant

Figure 7. Close-up of the fracture to corner joints to Component A (a) and to Component
CX (b). NP = Neutral Point.
The remaining solder balls on the PCB were broken away at some locations by gripping them
with a micro-tweezer and pulling vertically to the board surface. This enabled us to examine
the integrity of the joints towards the board pads. In all cases, the pads were ripped off from
the laminate. For a connection in the inner row, no fracture in the laminate could be observed

5
(Fig. 8) whereas extensive cracking was observed for corner joints (Fig. 9). The cracks had
formed on the inward side of the joints. For one solder joint, the “inside” of the via-in-pad
was coloured red indicating damages to the copper plating in the via hole (Fig. 10). The
results were similar for Component CX (Figs. 11-13).

Figure 8. Views showing the fracture between a pad and the board laminate for a ball in the
inner row to Component A. View (a) shows the ball with the underside of the
ripped-off pad and view (b) shows the location on the PCB where the ball was
attached

Figure 9. Views showing the fracture between a pad and the board laminate for a corner
ball to Component A. View (a) shows the ball with the underside of the ripped off
pad and view (b) shows the location on the PCB where the ball was attached

6
Figure 10. Views showing the fracture between a pad and the board laminate for a corner
ball to Component A. View (a) shows the ball with the underside of the ripped off
pad and view (b) shows the location on the PCB where the ball was attached

Figure 11. Views showing the fracture between a pad and the board laminate for a ball in the
inner row to Component CX. View (a) shows the ball with the underside of the
ripped off pad and view (b) shows the location on the PCB where the ball was
attached

Figure 12. Views showing the fracture between a pad and the board laminate for a corner
ball to Component A. View (a) shows the ball with the underside of the ripped off
pad and view (b) shows the location on the PCB where the ball was attached

7
Figure 13. Views showing the fracture between a pad and the board laminate for a corner
ball to Component A. View (a) shows the ball with the underside of the ripped off
pad and view (b) shows the location on the PCB where the ball was attached.

References
1 S. C. Bolton, A. J. Mawer, and E. Mammo, Influence of Plastic Ball Grid Array
Design/Materials Upon Solder Joint Reliability, The International Journal of
Microcircuits and Electronic Packaging, Vol. 18, No. 2 , 1995, pp. 109-120.
2 P.-E. Tegehall and B. Dunn, Assessment of the Reliability of Solder Joints to Ball and
Column Grid Array Packages for Space Applications, ESA STM-266, ESA
Publications Division, Noordwijk, 2001 (for ordering information see
http://esapub.esrin.esa.it/publicat/epdgi.htm)..